AI is changing who has power at work

I met for coffee with an engineering manager last week, and we were talking about a fascinating experience we both had made when it comes to AI at work.

She shared with me that she used to think when she was sitting in meetings about AI, everyone else understood something she didn’t, and that she was the only one in the dark. Many people would use acronyms and technical terms she didn’t know, and they sounded so sure of themselves. They seemed to have all the knowledge and power, and she felt like she knew nothing.

Then she took a short course on AI, to understand the business side of it. And she realised that when you really dig in, a lot of those confident people don’t understand it either. They feel threatened by it. And they use jargon to cover up the fact that they’re unsure.

That conversation made me think. How is AI changing power dynamics at work? What does that mean for us?

AI is shifting power

Back in the 1950s, two researchers, John French and Bertram Raven, studied where power at work comes from. They found a handful of sources. One was very simple: if you know something other people need, and they can’t easily get it anywhere else, you hold power over them. They called it expert power. (I have linked the research below)

Think of the one person in your office who knows how a legacy system works. Everyone has to go through them. That’s expert power, and it comes from having knowledge that’s hard to find.

What is happening with AI coming into work?

A lot of knowledge that used to be locked inside a few experts’ heads is now just a question away. Need to understand something technical? You can ask Claude or ChatGPT. Your company might even have its own knowledge base you can chat to, instead of chasing down the one person who knows.

When knowledge becomes easy to get, the people who held it lose some of their power. The expert everyone used to depend on simply isn’t needed as much anymore.

It’s happening to managers too. Many managers built their influence on owning a process or a decision. But more and more of those decisions are now handled, or at least shaped, by AI. The ground is shifting under them as well.

There’s fascinating research from Harvard about this: It found that managers often feel so threatened by AI that they quietly block it, even when it would help the business. This is called “the Manager’s AI Dilemma”. (article linked below) One finding stayed with me: the fear of losing status can be even stronger than the fear of losing your job.

Why some people hide behind jargon

So who is gaining power now? The technical experts - and anyone who understands even a little about AI. And that creates a strange side effect.

Some people who feel their old power decrease, try to build it in a new way. They pick up the AI words and throw them around, hoping to sound like a technical expert. Often they’re not trying to teach you anything at all. They’re trying to feel bigger by making you feel smaller.

Once you see this, everything changes. The jargon stops sounding like proof that someone is ahead of you. Often it’s a sign that someone is worried about being left behind.

So what do you do the next time you’re sitting in one of those meetings? Here are three things.

1. See what’s really going on

When someone starts talking jargon, remember this: There’s a good chance their old source of influence is eroding, and the jargon is now what they are holding on.

Seeing this behaviour for what it is takes away its power over you. You stop assuming they know something you don’t, and that changes how you show up.

2. Change the conversation to what matters

When the talk gets lost in technical detail, bring it back to what the business cares about. We often dive into the “how” of AI before anyone has asked the bigger questions. How does this help our customers? How much money does it make or save us?

You don’t need to talk technical jargon back - instead, change the level of the conversation. Try something like:

“This is useful detail. Can we zoom out for a second, though? What does this actually change for the customer?”

Notice what that does. You’re not giving the impression that you are evading the subject - instead, you are levelling the conversation up to the question that counts, which is exactly what a leader does.

If the person truly knows their subject, they’ll answer confidently, and now the two of you are talking as equals. If they don’t, your question quietly shows it, without you ever having to challenge them.

3. Get your own hands on the tools

You can’t think clearly about something you’ve never touched. So the best thing you can do is start using these tools yourself.

Let’s say your team is sitting down to discuss bringing in a tool like Claude Cowork, and you walk in having already tried it. You have your own experience to draw on.

And remember, you’re not trying to become an AI expert. Nobody can be one - the field moves far too fast for anyone to stay fully on top of it. You only need to have touched the tools.

So start small. Find your own use case for an AI tool that your team might be using at work - that use case can be from your private life, hobbies etc. E.g. I created a Claude project to help me write music concert programmes. Spend time to implement it yourself. That’s enough to share useful insight and contribute to the discussion at work.

This is a huge opportunity for you
 

Here’s what the engineering manager and I both ended up on over that coffee. The knowledge that used to set a few people apart is becoming something anyone can reach. When everything is moving like that, the calm person who asks the clear question is the one who comes across as a leader - even more so when you’ve already tried the tools yourself.

So don’t wait for anyone to bring you in. Pick one AI tool your team is considering, spend an afternoon with it this week, and bring that experience into your next meeting. Power is shifting - use it to your advantage.

PS. The research on power I mentioned is linked here. The Harvard article on the Manager’s AI dilemma you can find here.